Record

03

Company Three

Treasury reporting, rewritten for people who hate it

A finance product where the hardest problem turned out to be language, not layout. Clients were phoning their manager to ask what their dashboard meant.

Diagram counting the screens in a treasury reporting tool, with a span of retired screens marked out.

Sector

Banking

Role

Design lead

Years

2024–25

Constraint

Regulated copy. Every label cleared by legal.

The brief was wrong. Almadine asked for a clearer dashboard. What they had was a vocabulary problem wearing a layout problem's clothes.

We started in the call logs, not the product. Six weeks of support transcripts gave us the hundred questions clients actually ask, in the words they actually use — and that list became the specification.

Position summary table with labelled columns, showing figures rewritten in plain language.
Position summary table with labelled columns, showing figures rewritten in plain language.

Position summary, after

Decision log

Four that mattered

Wrote the glossary first

No screen was drawn until a hundred and forty terms were agreed with legal and the desk. Design review stopped being a copy argument. It also put five weeks between kickoff and the first visible artefact, which is a long time to ask a client to trust you.


Deleted nine screens

Every screen had an owner who wanted it kept. Nine went, and the largest measured gain in the project came from subtraction rather than any new feature. Two of those owners lost the only surface their team was visible on, and I handled that conversation badly the first time.


One number per view

Clients asked one question at a time; the product answered six at once. Each view was cut to a single number, with everything secondary moved behind a deliberate second action. The relationship managers lost their dense overview and objected loudly for a quarter.


Tested against the phone script

Labels were run against transcripts of real support calls rather than a usability lab. If a label did not answer the question a client actually asked, it was rewritten. It is a slow loop, roughly two weeks a round, and it only works where a support archive already exists.

Outcome

Measured twelve weeks after launch

62%

Fewer 'what this mean' calls

140

Labels rewritten and cleared

4.6

Client satisfaction, from 3.1

What I would do differently

The part most portfolios leave out

Legal review was treated as an end-of-line gate for the first two months and it cost a full sprint. Bringing that reviewer into weekly design sessions turned the slowest dependency into the fastest one.

Create a free website with Framer, the website builder loved by startups, designers and agencies.